# interpret logit vs fixed effects logit (conditional logit) estimates

I'm running two logit models on passing or failing an exam - one random effects logit, and one fixed effects logit (conditional logit) where I use "community" as my group variable. I've set my estimates to be log-odds ratios in both models. My question is, is it sensible to compare estimates from these two models or am I comparing apples and oranges?

• Clarification: ¿what do you want to compare between the two models? In particular, ¿are you looking to just compare the parameter estimates? or ¿do you want to compare predicted values? Commented Mar 24, 2018 at 13:29
• Im just looking to compare odds ratio estimates :-) For example, I get a log-odds ratio for the effect of "age" on passing the exam of 1.06 in the std. logit and 1.08 in the community fixed effects logit Commented Mar 24, 2018 at 14:04
• sorry...one more clarification: when you say random effects vs. fixed effects (conditional), ¿are you using the groupings (community) as random collections vs. fixed effects for each specific group? I.e., if you have 10 groups, one model just estimates a random effect for the groups (generally), but the other model has a dummy variable for each specific group (less the reference group). Commented Mar 24, 2018 at 15:46
• No problem :-) the random effects is just a logistic regression logodds = intercept+ beta_1*x1 + ... beta_n*xn where the explanatory variables are both continous and dummy. The fixed effects logit is within the "panel data econometrics" tradition, such that it estimates the within-group relationship between the independent variables and the binary dependent variable. Commented Mar 24, 2018 at 16:19
• Sorry to belabor this, but I'm not sure this answers my question in full. The inclusion of panel (e.g., group-level means) requires the inclusion of k-1 dummy variables for k groups (and each group then has an estimate); what is unclear to me when you use the term "random effects" is ¿what parameter are you treating as random? My inclination is that you are treating the intercept as random. Commented Mar 24, 2018 at 18:03